3 Steps to Reverse Engineering

The ultimate goal of reverse engineering is generally to create a 3D CAD (Computer Aided Design) model for use in the customer’s design system. There are several types of models that can be generated from 3D scanned data and a large varietyof reverse engineering software to choose from. Below is a simple guide to the complete reverse engineering process.

  1. Determine Exactly What You Want

Before the process even begins, the intent must be determined. Users must ask themselves a series of questions in order to figure out just exactly what they hope to achieve. Some of these questions include:

  • How will this image be used?

Knowing exactly what a project’s end-use will be is a vital part of the reverse engineering process. In order to achieve optimal results in the most efficient manner, this must be determined. Otherwise, the reverse engineering process can become very time-consuming and rather inefficient. For example, someone in the automotive industry mightask for the scanning of an entire engine, when all they hope to do is determine fit and clearance. If you were unaware of what they were setting out to obtain, the reverse engineering process would be extremely lengthy in order to capture and re-create every little component of the engine. However, by asking the simple question “what is it going to be used for?” you would find that the process can be greatly simplified because an in-depth volumetric representation – and lengthy reverse engineering process – isn’t necessarily even needed.

  • Do I anticipate any changes?

If change is anticipated, the methodology used completely changes. Parametric CAD features are organized and structured within a "tree, " or programmed process to which the software assembles the model. When planning changes to the model, care must be taken during the modeling process to allow the easy modification or addition of features within this tree without confusing the software, and consequently destroying the model.

If no changes from the scanned part are planned, or the customer specifies they only require an IGES or STEP file, the engineer may move forward with modeling as he or she pleases without care to downstream alterations.

Greg Groth of Exact Metrology explains how and why the process alters if change is expected:

“Related to this is another common request I receive from customers: ‘I need a Parametric IGES or STEP file.’ This is a contradiction. Parametric CAD models contain design intent, a structured combination of prismatic 3D features driven by specific dimensions. An IGES or STEP output is a stripped version of that parametric 3D CAD model, typically represented by that outside “skin” of the CAD surface, allowing users to share 3D data between multiple platforms that do not share a common internal language.”

  • What are your tolerance requirements?

Knowing how accurate the part needs to be is another crucial element of the reverse engineering process. 3D scanners, such as Creaform's HandySCAN 700, provide metrology-grade measurements with accuracy up to .001". Often times, a level of accuracy this high may not even by necessary, which will considerably shorten the modeling process.

Why would we not always strive to achieve a model that is perfectly accurate to the scan data? The manufacturing process creates variance between parts, and therefore acceptable tolerances must be designated for each feature of a part. A perfect measurement on one component may not be representative of the next one down the line.

Let us consider a sand casting. A majority of the outer surface will likely have a very large tolerance window. Often times the surface will be rougher than the tolerance achieved by the scanner, contain defects, and part-to-part variance will be high. Trying to re-create this surface to 0.001" would likely be a large waste of valuable time.

Now consider the inverse, which may be a machined mating surface on that casting, performed in a secondary manufacturing operation. This is likely to be much more controlled, and thus require much greater accuracy. The engineer would be wise to create this geometry with more care.

  1. Data Acquisition

3D scanning technologies have advanced greatly over recent years, and selecting the appropriate type of 3D scanning technology for a project is extremely important. To determine this, the following questions must be answered:

•Which 3D Scanning Technology to Choose?

Before selecting which 3D scanner we would like to go with, we must understand the capabilities and limitations of each technology. Not one covers every area of application.

  • Laser: extremely fast and capable of scanning various surface finishes, useful in scanning a variety of objects.
  • Structured Light: provides extremely precise measurements for small objects and those of delicacy.
  • Phase Shift/Long-Range: capable of obtaining proportionally accurate measurements of extremely large objects from extended distances
  • Contact/Portable CMM: primarily useful in quality control. Very accurate method of defining simple geometric features. Can potentially deform or destroy original object.

As you can see from the previously discussed types of technologies that are listed above, the type of object is a huge component in determining what type of 3D scanner best suits your need.

•Where Are You Scanning?

You must first considerwhereyou will be scanning in order to determine whether you’ll need to select a stationary 3D scanner or a portable one. Will it be left strictly in a metrology room with controlled temperature and vibration? A tripod-based structured light unit works great in these conditions. If parts are being brought into a room, it's likely they are small, and will easily fit within this technology's scan volume.

Will you be scanning on an active shop floor? You'll need a more robust unit that is easily transportable, and largely unaffected by the environment. A Creaform laser scanner is ideal in this situation, as it can dynamically track the part while under the same motion that would wreak havoc on a structured-light unit.

What about outside, in direct sunlight? This all but eliminates any structured-light technology. Laser scanners work well if shade or a canopy is available, however long-range scanners have no issue with the sun and are the best choice if the part size allows. Coming round full circle, a long range scanner would fare poorly attempting to define fine detail on small parts.

•How Big Is Your Object?

Dimensions are the next determinant that needs to be addressed in selecting the appropriate 3D scanner. 3D laser scanners are primarily useful in scanning moderate-to-large objects. Long-range scannersare best suited to very large objects/structures, and area scans. Opposite to this, structured light scanners are highly effective at scanning small, highly-detailed items. If object preservation is not a factor, the item is not of great size, and simple geometric features are all that is required, you could use a contact portable CMM as well.

If we consider a vehicle as an example, a laser scanner would be ideally suited to scan most interior and exterior components, panels , and geometry, up-to and including the entire vehicle. A structured light scanner would be very useful for scanning the vehicle's smaller individual components and assemblies (think A/C vents, clips, connectors, etc.) which all require very fine details for accurate re-creation. A portable CMM may be the best technology to define component mounting locations on the vehicle's frame, and a long-range scanner would be the choice for scanning the floor plan of the entire manufacturing facility.

•What Is The Surface of the Object?

The external surface of an object is a major factor is choosing the best3D scanner for a job as well. The material can potentially cause major problems if the appropriate scanner is not used. For example, reflective surfaces can cause deformity in the data mesh ofstructured light technologies, whereas laser-based systems have very nearly made this obstacle a thing of the past. One solution to adverse surface material can be to apply an opaque layer to the object when scanning, if you are biased to one type of scanning technology.

•How High Do I Want the Resolution to be?

Depending on what your use for a 3D scan is, the resolution may be one of the mostpivotal attributes needed from the scan.

First, we must clarify that accuracy and resolution are two separate values. In the 3D scanning world, accuracy is the measure of how close your scanned data is to the actual part, whereas resolution defines the size of the feature that we can capture.

If you are in need of a highly accurate scan, you’ll more than likely be spending significantly more time 3D scanning. The resultant file would be exponentially larger, thus requiring more time and processing power for the reverse engineering process.

If we re-visit the car example, you would not need a mesh of the same resolutionto define the exterior geometry of the car as you would of an A/C vent sub-assembly. It would simply take too much time to scan, and the resulting file size of the mesh would make the reverse engineering process a nightmare. However, if we consider the re-creation of the vent itself, this higher resolution would help to more clearly and accurately define its small features.

This takes us back to the question of, what are you using the scans for? Once you are able to answer this, how much resolution you need to complete the job should be more easily determined.

  1. Data Processing

Now that we've determined what we will be scanning, and which scanner is appropriate for the situation, we must decide how we will be processing the data.

Knowing your equipment's capabilities and weaknesses, and how to most effectively use it, is extremely important. The quality of data gathered during the scanning session will directly impact the time spent in downstream processing. It's very difficult and time consuming to create a quality model from bad scan data, meaning skipping a step or two that may save you minutes during data acquisition could lead to hours of additional time while reverse engineering.

Additionally, make sure you have captured enough data to properly align components of an assembly, or accurately piece together a large data set that was too much to grab in one scanning session file. Going into the project with a clear plan will save you from the eventual choice of making an assumption during the modeling process, or returning to capture more data that was missed.

How will you create the polygonal mesh file from your captured point cloud? Many professional 3D scanners come with their own software to provide this file type as a direct output. If this is not the case with your equipment, there are many mesh manipulation software packages on the market to make this conversion.

How do we go about generating the CAD model? In days past, this was a very clunky endeavor. The engineer was forced to use CAD not initially designed for the application to create crude 3D cross sections through the mesh, convert those to useable parametric features, all while battling the software as its processing power wasstrained by its interpretation of the imported mesh.

Today, advanced software options, like Geomagic, are optimized specifically for reverse engineering. Geomagic Design X, specifically, has the ability to accept data directly from a 3D scanner, generate a high resolution mesh, then separate that mesh into useable regions to be used for instant creation of individual parametric features. It also has a NURBS surface wrap option to lay a highly accurate and detailed surface directly over extremely complex geometry.

The final step to reverse engineering is verification. How do we know the part we modeled matches the part we scanned? It starts with choosing the appropriate 3D scanner for the application, and ends with a comparison of the model to the scan data. Once a model has been generated, the software often contains a comparison tool allowing the user to set a tolerance band, and verify that much of their model falls within an acceptable deviation range from the scan data. This is often represented by a color map laid across the model, with each color representing a distance from the model surface to its accompanying vertex of the mesh.

Once the 3D model has been verified, it is then ready to be sent to the end user for manufacture.

Original Article: